hive-mcp-openclaw-bridge
MCP server that wraps Hive Gamification's 8 BOGO doors behind a 4-level hierarchical budget with atomic reservation, upstream POST, and settlement/release, implementing OpenClaw's 3-layer payment governance.
README
hive-mcp-openclaw-bridge
Reference integration with @kinthaiofficial's OpenClaw 3-layer payment governance framework. MCP server that wraps Hive Gamification's 8 BOGO doors behind a 4-level hierarchical budget — atomic reserve, upstream POST, settle/release. Real rails. No mock settlement.
Brand: Hive Civilization gold #C08D23 (Pantone 1245 C).
License: MIT.
Why this exists
OpenClaw's three-layer model splits payment governance into three concerns:
| Layer | Concern | Where it lives |
|---|---|---|
| Layer 1 | Internal hierarchical budget — atomic reserve / settle | this server |
| Layer 2 | Inter-agent millicent ledger settled at epoch boundaries | OpenClaw |
| Layer 3 | External x402 settlement on real rails | Hive Gamification |
This bridge demonstrates the Layer 1 ↔ Layer 3 seam. Every paid call to a Hive door (computational-debt, protocol-consensus-tax, pheromone-priority, zk-compliance-shield, capability-leasing, guild-synergy, audit-trail-premium, bounty-routing) is gated by an atomic reservation against every level in the caller's scope path:
Namespace → User → Agent → Task
The most-restrictive level governs. If any level is short of headroom, the upstream HTTP request is never made and no x402 settlement is signed.
The same BudgetHook interface is exposed from @hivemorph/qvac-client v0.2 — drop your own implementation in to wire this hierarchy into a different runtime.
Background reading
- Agent wallet economic models for autonomous agents — kinthai
- 221 agents: multi-agent coordination lessons — kinthai
- agents.kinthai.ai — running on OpenClaw
Tools
| Tool | What it does |
|---|---|
budget_status |
Inspect caps, spend, reservations, headroom across all 4 levels |
budget_set_cap |
Override a per-key cap at one level |
budget_check |
Read-only pre-flight against a proposed amount + scope |
doors_list |
List all 8 BOGO doors with endpoint, price, asset, term |
door_economics |
Fetch live /economics for one door |
bogo_claim |
Claim first-use-free via /v1/bogo/claim |
door_subscribe |
Canonical gated path — atomic reserve → upstream POST → settle/release |
Doors
| Door | Endpoint | Price | Term |
|---|---|---|---|
| computational-debt | /v1/debt/subscribe |
0.50 USDC | 30d |
| protocol-consensus-tax | /v1/pct/subscribe |
0.25 USDC | 1 read |
| pheromone-priority | /v1/pheromone/subscribe |
5.00 USDC | 30d |
| zk-compliance-shield | /v1/compliance/subscribe |
1.00 USDC | 30d |
| capability-leasing | /v1/lease/start |
0.15 USDC | 1h |
| guild-synergy | /v1/guild_synergy/subscribe |
10.00 USDC | 30d |
| audit-trail-premium | /v1/audit_premium/subscribe |
3.00 USDC | 30d |
| bounty-routing | /v1/bounty_routing/route |
0.50 USDC | 1 routing |
Backend: https://hive-gamification.onrender.com. All endpoints are live and verified.
Lifecycle of a paid call
client → tools/call door_subscribe { door_id, did, scope }
→ atomicReserve(scope, price, asset)
├─ each level checked: namespace, user, agent, task
├─ headroom = cap − spent − reserved
└─ short on any level? reject before signing
→ POST hive-gamification.onrender.com<endpoint>
→ 2xx? settle(reservationId)
non-2xx or throw? release(reservationId)
→ return { ok, reservationId, binding_level, upstream }
Quickstart
git clone https://github.com/srotzin/hive-mcp-openclaw-bridge
cd hive-mcp-openclaw-bridge
npm install
npm start
Then from any MCP client:
POST http://localhost:3000/mcp
{ "jsonrpc": "2.0", "id": 1, "method": "tools/list" }
Configuration
| Env | Default | Meaning |
|---|---|---|
PORT |
3000 |
listen port |
HIVE_BASE |
https://hive-gamification.onrender.com |
upstream base URL |
HIVE_TIMEOUT_MS |
15000 |
upstream fetch timeout |
NAMESPACE_CAP |
50 |
default namespace-level cap (USDC) |
USER_CAP |
10 |
default user-level cap |
AGENT_CAP |
2 |
default agent-level cap |
TASK_CAP |
0.50 |
default task-level cap |
Per-key caps override defaults via budget_set_cap.
Related
- @hivemorph/qvac-client v0.2 — TypeScript SDK exposing the same
BudgetHookinterface against any x402-enabled endpoint. - Hive Gamification agent card — A2A 0.1 advertisement of the 8 doors.
License
MIT — see LICENSE.
Hive Civilization Directory
Part of the Hive Civilization — agent-native financial infrastructure.
- Endpoint Directory: https://thehiveryiq.com
- Live Leaderboard: https://hive-a2amev.onrender.com/leaderboard
- Revenue Dashboard: https://hivemine-dashboard.onrender.com
- Other MCP Servers: https://github.com/srotzin?tab=repositories&q=hive-mcp
Brand: #C08D23 <!-- /hive-footer -->
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。